# AgentIF: instruction following in agentic scenarios (AgentIF)

> Instruction-following benchmark built from real agentic applications, with long instructions that carry many constraints.

Canonical page: https://benchlm.ai/benchmarks/agentif

- Category: [Instruction Following](/instruction-following)
- Last updated: October 29, 2026

## About AgentIF

- Year: 2025
- Tasks: Agentic system prompts and instructions
- Format: Instruction success rate
- Difficulty: Complex agentic instructions
- Paper: [AgentIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios](https://agentif.github.io/)

NVIDIA reports it for Nemotron 3.5 Bolt E4B (October 23, 2026), run in its NeMo Gym harness, as the instruction success rate in reasoning mode.

AgentIF is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (2 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Nemotron 3.5 Bolt E4B BF16](/models/nemotron-3-5-bolt-e4b-bf16) | NVIDIA | 25.81% |
| 2 | [Nemotron 3.5 Bolt E4B](/models/nemotron-3-5-bolt-e4b) | NVIDIA | 22.95% |

## FAQ

### What does AgentIF measure?

Instruction-following benchmark built from real agentic applications, with long instructions that carry many constraints.

### Which model scores highest on AgentIF?

Nemotron 3.5 Bolt E4B BF16 by NVIDIA currently leads with a score of 25.81% on AgentIF.

### How many models are evaluated on AgentIF?

2 AI models have published results on AgentIF in the BenchLM catalog.

### Does AgentIF affect BenchLM's overall score?

Not directly. AgentIF is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Compare Top Models on AgentIF

- [Nemotron 3.5 Bolt E4B BF16 vs Nemotron 3.5 Bolt E4B](/compare/nemotron-3-5-bolt-e4b-vs-nemotron-3-5-bolt-e4b-bf16)